Effective Rate Analysis of MISO Systems over $\alpha$-$\mu$ Fading Channels

arXiv: 1507.04243v 1 [cs.IT] 15Jul2015

l

Effective Rate Analysis of MISO Systems over a-/i

Fading Channels

Jiayi Zhang*!, Linglong Dai*, Zhaocheng Wang*, Derrick Wing Kwan Ngt, and Wolfgang H. Gcrstackch
*Tsinghua National Laboratory for Information Science and Technology (TNList)

Department of Electronic Engineering, Tsinghua University, Beijing 100084, P. R. China
i Institute for Digital Communications, University of Erlangen-Nurnberg, D-91058 Erlangen, Germany
Email: {jiayizhang, daill, zcwang}@tsinghua.edu.cn, {kwan, gersta}@lnt.de

Abstract —The effective rate is an important performance
metric of real-time applications in next generation wireless
networks. In this paper, we present an analysis of the effective
rate of multiple-input single-output (MISO) systems over a-
p fading channels under a maximum delay constraint. More
specifically, novel and highly accurate closed-form approximate
expressions of the effective rate are derived for such systems
assuming the generalized 0-// channel model. In order to examine
the impact of system and channel parameters on the effective
rate, we also derive closed-form expressions of the effective
rate in asymptotically high and low signal-to-noise ratio (SNR)
regimes. Furthermore, connections between our derived results
and existing results from the literature are revealed for the sake
of completeness. Our results demonstrate that the effective rate is
a monotonically increasing function of channel fading parameters
a and //, as well as the number of transmit antennas, while it
decreases to zero when the delay constraint becomes stringent.

Index Terms — Effective rate, generalized a-// fading channel,
multiple-input single-output (MISO), QoS provisioning.

I. Introduction

Shannon’s ergodic capacity of multiple-antenna systems
has been extensively analyzed in pioneering works 0, 0-
However, it cannot account for the quality of service (QoS)
requirements of some emerging real-time applications for next-
generation wireless networks, such as mobile video phone,
interactive gaming, and multimedia streaming, where the tol¬
erable delay is limited. Therefore, a novel performance metric
is required to provision delay guarantees for such real-time
applications. Motivated by this fact, Wu and Negi El proposed
the concept of effective rate (or effective capacity, effective
throughput) to consider the effect of statistical delay QoS
guarantees for the achievable transmission rate. More specif¬
ically, the effective rate is defined as the maximum constant
arrival rate at the transmitter when guaranteed statistical delay
constraints can be satisfied.

Recently, the concept of effective rate has attracted many re¬
search interests for the analysis of single- and multiple-antenna
communication systems El-0- The effective rate of wireless
networks was investigated for adaptive modulation and coding

This work was supported in part by the International Science & Technology
Cooperation Program of China (Grant No. 2015DFG12760), the National
Natural Science Foundation of China (Grant No. 61201185 and 61271266),
the Beijing Natural Science Foundation (Grant No. 4142027), the Foundation
of Shenzhen government, and China Postdoctoral Science Foundation (No.
2014M560081).

at the physical layer with an automatic repeat request protocol
at the data-link layer in El- In 0, an analytical model
of the effective rate for proportional fair scheduling used
in orthogonal frequency-division multiple access (OFDMA)
systems was presented. For multiple-antenna systems, the
authors in E) proposed the optimal power allocation scheme
with statistical QoS provisioning to maximize the effective
rate of virtual multiple-antenna wireless networks. The per¬
formance of multiple-antenna systems under QoS constraints
is captured in 0 through the effective rate formulation.
Due to its low complexity and high performance, the MISO
system is a promising technology for next-generation wireless
networks. Therefore, a plethora of recent works focused on the
effective rate of multiple-input single-output (MISO) systems,
which cover the analysis of independent 0 and correlated
0-031 fading channels. However, these presented results
are based on the assumption of a homogeneous scattering
environment corresponding to Rayleigh, Rician, Nakagami-
m, or generalized-/\ fading models. In practice, the received
signals reflected by a surface are spatially correlated, and the
resulting signal is obtained in a nonhomogeneous scattering
environment modeled by a nonlinear function of the modulus
of the sum of the multipath components. To this end, the a-
// fading channel is proposed in lfl6l to better accommodate
the statistical variations of the propagated signal in diverse
field measurements of propagation environments fl2l . ffTJl . It
has been proved that the o-fi fading model includes Rayleigh,
one-sided Gaussian, Weibull, exponential, Nakagami-m, and
Gamma models as special cases. Against this background, a
basic understanding of the effects of nonhomogeneous o-fi
fading on the effective rate performance of MISO systems is
provided in this paper. Only recently, we have derived novel
expressions for two nonhomogeneous cases: 77 -p and K-fi
fading channels d, Q3). To the best of authors’ knowledge,
however, there is no literature on the effective rate of MISO
systems over a -// fading channels. The reason behinds this
is twofold: First, the derivation of the exact expression for
the probability density function (PDF) of the sum of a-fi
statistics is known to be very challenging; Second, significant
mathematical challenges for the effective rate analysis are
incurred when considering the complicated generalized a-fi
fading.

In this paper, we derive highly accurate closed-form expres¬
sions of the effective rate of MISO systems over independent

2

and identical distributed (i.i.d.) a-p fading channels. Note
that these expressions are given in the form of Meijer’s-G
functions 01 and Fox’s -H functions DU, which can be
easily evaluated and efficiently programmed in Mathematic a.
In order to intuitively investigate the impact of system and
channel parameters on the effective rate, we also derive closed-
form expressions of the effective rate in the asymptotically
high and low signal-to-noise ratio (SNR) regime, respectively.
Our results prove that the effective rate increases with larger
value of fading parameters a and p, as well as more transmit
antennas, while the tightened delay constraint will reduce
the effective rate. For the sake of completeness, connections
between our derived results and previously presented results
are also provided in this paper.

The rest of the paper is organized as follows. Section [IT]
describes the system model and the statistical characteristics
of generalized a-p fading channels. In Section!©] we derive a
novel expression for the exact effective rate of MISO systems
over i.i.d. o- p fading channels. To quantify the effect of
system and channel parameters on the effective rate, we also
derive closed-form expressions both for high SNRs and for
the minimum transmit energy per information bit. Finally,
theoretical and simulation results are compared in Section [IV]
while Section [V] concludes the paper.

Notations: Vectors and matrices are represented by lower¬
case bold typeface and uppercase bold typeface letters, respec¬
tively. The symbol E{-} refers to the expectation operator of a
random variable, Pr[ ] accounts for the probability, and tr (•) is
the trace of a matrix. We use (-)t to represent the Hermitian
transpose, and the natural logarithm of a number is defined
as ln(-). Moreover, I denotes the unit matrix. Finally, G[-]
represents the Meijer’s-G function jT8j Eq. (9.301)] and H[-]
stands for the Fox’s -H function 119], Eq. (2.1)].

II. System and Channel Model
A. System Model

We consider a point-to-point MISO system, where the trans¬
mitter consists of N t antennas and the receiver is deployed
with single antenna. The input-output relationship over block
fading channels can be expressed as

y = hx + n, (1)

where h G C lxNt denotes the MISO channel’s fading vector
with elements h = [hi, ■ ■ ■ , hk, ■ ■ • , x € C^** 1 repre¬
sents the transmit signal vector with the covariance E{xx'} =
Q subject to the constraint tr(Q) < P, where P denotes
the transmit power. Moreover, n represents complex additive
white Gaussian noise (AWGN) whose samples are i.i.d. zero-
mean complex Gaussian random variables with variance Nq.
Furthermore, we assume that the instantaneous channel state
information is not available at the transmitter and equal power
is allocated to each transmit antenna, i.e., Q = lj-I, so the
transmit SNR is p =

transmitter is stationary. Then, the effective rate of the service
process is defined as

« ( 0) = -^ ln(E{exp( -^G )}) , 9^0, (2)

where G represents the random throughput during a single
block and T denotes the block length. The delay QoS exponent
9 is given by

9 = — lim

Zfh —yoo

hi (Pr \L > Zth])

Gi

(3)

where / th denotes the specified threshold of queue length and
L is the equilibrium queue-length of the buffer assumed to be
available at the transmitter. Note that we use the queue length
as a QoS performance metric instead of delay to obtain an
intuitive and tractable analysis. Moreover, the probability of
L > decreases faster with a larger 9. On the contrary, the
system can tolerate an arbitrarily long delay when there is no
delay constraint as 9 —t 0. Capitalizing on such conditions, the
effective rate coincides with the classic concept of Shannon’s
ergodic capacity.

In the case that the transmitter sends uncorrelated circularly
symmetric zero-mean complex Gaussian signals, the effective
rate of the MISO channel can be succinctly expressed as II

= 0g 2 j (i + hht ) bit/s/Hz,

(4)

where A = with B denoting the bandwidth of the

system. In particular, A represents a metric of delay constraint.
It is clear from © that the effective rate depends on the
distribution of fill' , which is analyzed in the following.

C. Sum of a-p RVs

The a-p distribution is a general fading model, which
accounts for the nonlinearity of a nonhomogeneous propa¬
gation environment. It consists of two parameters describing
the physical properties of the fading, namely a and /i. More
specifically, the power parameter a represents the nonlinear
function of the modulus of the sum of the multipath com¬
ponents, while the parameter p is related to the number of
multipath clusters. For a single a-p fading link, the PDF of
the instantaneous received SNR is given by El Eq. (8)]

/71 (7l)

ai7i“ lMl / 2 1

-To-exp

2 ff w/ 2 r(/ti)

(5)

where fii = E { 71 } T (pf) /V (p\ + 2/ai), E{ 7 i} =

f 2 r (pi + 2 /ai)/(/ri 2 /“ i r (pi)), and f\ is defined as the en¬
root mean value of the envelope random variable R, i.e.,
fi = a \/E {f? ai }. Using lfl8l Eq. (8.326.2)], the ?ith moment
of 71 can be expressed as

E{7i n }

( 2 n

( hi H-

V £*1

( 6 )

B. Effective Rate

As introduced in 0 , we suppose that the data arrives in
the buffer at a constant rate, and the service process of the

Let 7 = E^i 7 , denote the sum of N t i.i.d. a-p branches
instantaneous SNRs. It was shown in El that a single a-p
RV can exploited to approximate the sum of a-p RVs, and

3

this approximation seems to achieve accurate results. More
specifically, moment-based estimators are used to calculate the
parameters a, //, and f from the exact moments of 7 . To this
end, we need to derive the parameters a, p, and f by solving
the following nonlinear equations

E 2 (7) _ ^(p+l/a)

E ( 7 2 ) — E 2 ( 7 ) T (p) T (p + 2/a) — T 2 (p + 1/a) ’

E 2 (7 2 ) T 2 (p + 2/a)

E ( 7 4 ) - E 2 ( 7 2 ) r (p) r (p + 4/a) — T 2 (p + 2/a) ’

= M 1/a r (p) E ( 7 )

T(M+l/a) ’ U

where the moments in © can be evaluated by using the
multinomial identity |20] Eq. (10)]:

0N t -2

q

h

E(7*)=E - E

j 1=0 3N t -1=0

xE( 7 r i )-E( 7 ^ 1 ).

jNt — 2 \

jN t -1 /

( 8 )

Note that an analytical solution to the system equations © is
very difficult to obtain, while we can use numerical methods
instead, such as the fsolve function of Matlab and Maple.
Having obtained these parameters, the approximated PDF
expression of the sum can be written as

fi (7)

2/3°W 2 r (^) exp

(9)

III. Effective Rate over a-p Fading Channels

In this section, an exact an alysi© of effective rate of MISO
systems over i.i.d. a-p fading channels is presented. The
asymptotically high and low-SNR regimes are also considered
in order to provide physical insights into the impact of the
system and channel parameters on the effective rate.

A. Exact Analysis

Substituting © into © and using ll2Tl Eq. ( 8 .4.2.5)], Il22l
Eq. (11)] and ll22l Eq. (21)], we derive the exact effective
rate of MISO systems over a-// fading channels as (1 1 01 1 at the
bottom of this page, where A (e,r) = — ' , r+ *~ 1 ,

with r being an arbitrary real value and e a positive integer.
Moreover, l/k = a/2, where l and k are both positive integers.
For example, if a = 0.8, we should set l = 2 and k = 5.
Furthermore, l = a and k = 2 for the special case of a € Z+.
Note that the evaluation of (ITOl) can be done efficiently for
special values of a. However, for large values of l and k, it
is not efficient to compute (ITOt . Therefore, another method is
adopted in the following to overcome this problem.

'The term exact analysis is used to represent highly accurate approxima-
tions for the sake of distinguishing it from asymptotic analysis.

We recall the well-known translation from the Meijer’s G-
function to the Fox’s //-fuaction as ll2Tl Eq. (8.3.2.21)]

rrm,n

n p,q

X

[a p , 1]

r h n

_ pym,n

~ ^p,q

X

[Vp, -LJ J

01 )

With the help of [22] Eq. (11)] and [22] Eq. (21)], the power
and exponential functions can be expressed in terms of the
Fox’s //-fuaction as

= H,

1,0

(i+*r =

0,1

1

( 0 , 1 )

1,1

, and

(w + 1,1)
( 0 , 1 )

( 12 )

03)

respectively. Substituting (IT2l) and ( | 1 3[ i into ©. we get
the Mellin-Barnes integral of the product of two Fox’s II-
functions. Then, the effective rate expression can be written
as

R( Pl e)

1

A

- l°g 2

(r (A) r (/z))

- iog 2

(l,a/2)
CM), (A a/2)

(14)

where we have used the integral identity [21] Eq. (2.25.1.1)]
and the property ET1 Eq. ( 8 .3.2. 8 )] of Fox’s if-functions. It is
worth to mention that (ITdl) is very compact which simplifies
the mathematical algebraic manipulations encountered in the
effective rate analysis.

For the case of Nakagami-m fading channels, we set a = 2
and n = to. With the help of (23] Eq. (07.33.26.0004.01)],
the effective rate expression of (ITOt reduces to

R(P,6) = — ^ lo g2

ttp

mNt

1 / / mNt

- ^ioga [U (mN t ; mN t + 1 - A; -—

(15)

where U(-) is the Tricomi hypergeometric function l23l Eq.
(07.33.02.0001.01)] and O is the average gain of channel
fading. Note that (Q© coincides with © Eq. ( 6 )].

B. Asymptotic Analysis

In the high-SNR regime, we assume p is large in the initial
expression © to obtain the following asymptotic effective rate

R°° (p,0)

a(p/N t )~ A
2/3 q /V 2 T (/i)

X

^a.p./ 2 -A-l exp _

(16)

r(p,9)

1 1 f aVkl^iNtp/p) 1 *^ 2 \

^ z4° g2 ^ (27r) W-3/2 r(A)r(a) J

( s-ik-\-l,l
u l,k+l

(N t /p) l

_ (/3“/ 2 fc)

k

A (Z, 1 — ap/ 2)

A (/c, 0), A (Z, A — ap/2)

, ( 10 )

4

By invoking fl8l Eq. (3.326.2)] and after some straightforward
algebraic manipulations, we can derive a tractable result as

R°° (p,0)«log 2 (^

1,„ (V{p-2A/a)\

A log2 { r (M ) )

(17)

Note that the condition required for using |fl8l Eq. (3.326.2)]
for the integral in ( fl6l > is satisfied in our calculation by taking
A < a/i/2 — 1. The above result indicates that the high-
SNR slope is Soa = 1, which is independent of /3. The
same observations were made for the Rayleigh, Rician, and
Nakagami-m cases ©, 0.

We now investigate the effective rate in the low-SNR
regime. Intuitively, a second-order Taylor expansion for p —> 0
can approximate the low-SNR effective rate. However, the
authors in |24| proved that the Taylor expansion method may
in fact result in misleading conclusions regarding the impact
of the channel in the low SNR regime. Hence, it is beneficial
to explore the effective throughput at low SNRs as ll24l

R

Sc\og 2

(El/El \
W jVbnuJ

(18)

The normalized transmit energy per information bit ^ . and
the wideband slope So can be respectively formulated as

El

N 0t

= lim

p->o R (p, 9) R> (0, 9) ’
A 2[i?' (0, 0)] 2 In 2

So = -

R" (0, 0)

(19)

( 20 )

where R' (0,0) and R" (0,9) represent the first and second
order derivatives of R (0, 9) with respect to 9 , which can be
expressed as

R " (M) = nEi( E {hht} ) 2 “ i^tt2 E { (hht > 3 } ■

( 22 )

Recall that for i.i.d. MISO a-p fading channels,

E{hh»} = y)E{i/.,i 2 } = <23)

Using an approach similar to Appendix I of 0, we can derive

N t

N t N t

E {( hht ) 2 } = H e {i^i 4 }+Z! H e {m 2 n 2 }

k =1 k=l j=l,j^k

T 2 (m + 2/a)'

rw

■ (24)

Substituting (l23l >. ( 124| i. ( 12 1 | i. and < 122b into ( | 1 9l i and d20l) . the

minimum jA anc | the wideband slope So are respectively given
by (|25| > and ( [26b at the bottom of this page.

It is interesting to observe that the minimum jL is indepen¬
dent of the delay constraint A, whereas the wideband slope Sq
is independent of /3. Moreover, 5/ is a decreasing function in
A, while it is a monotonically increasing function in N t . These
results prove that a tighter delay constraint A decreases the
effective rate and an increasing number of antennas will yield
a higher effective rate, respectively. The maximum value of the
wideband slope is So = 2 when taking p —> oo or Nt —> oo.
Note that for the case of Nakagami-m fading channels (p = m
and a = 2), the minimum in (l25l) and wideband slope So
in (f26l) reduce to

El

N 0l

In 2

IT’

So =

and

2 mNt

(27)

(28)

A + 1 4- mN t ’

respectively, which is in agreement with ifTSl Eq. (20)-(21)].

IV. Numerical Results

In this section, we verify the analytical results presented in
Section [III] by computer simulations, and use them to study
the effective rate performance of MISO systems over a-p
fading channels. Without loss of generality, the bandwidth of
the system is normalized to B = 1 Hz. The simulation results
are derived by averaging the results over 10' i.i.d. a-p channel
realizations with unit power, which are generated by the sum
of independent Gaussian RVs method given in |16]. These
results provide meaningful insights regarding the impact of
different system and channel parameters on the effective rate
of MISO systems transmitting over a-p fading channels.

In Fig. Q] and Fig. [2] we consider the effects of fading
parameters (a and p) on the effective rate of MISO systems.
More specifically, the effective rate results of our computer
simulations are compared to that of the exact and high-SNR
approximate analytical expressions provided in (171)1) . (11 4l i. and
(1771) for different transmit SNRs. For comparison purposes,
we also show the effective rate of MISO systems over AWGN
channels as a benchmark. It is easy to see that the exact
analytical expression of the effective rate is very accurate in
the entire SNR regime. Besides, the high-SNR approximation
is quite tight even in moderate SNRs and its accuracy is
improved for larger values of the fading parameters, which
implies that it can efficiently predict the effective rate over
a wide range of SNR. Moreover, the high-SNR results show
that the high-SNR slope Soo is 1, which confirms our analysis
in Section IIII-BI Note that a performance improvement of
the effective rate is observed as a and p increases. This
observation is anticipated, since a large value of p results in

E b _ T (pi) In 2

No min /3lT (pi + 2/ai) ’

= _ 2W t r 2 (p + 2/a) _

0 {A + 1) (T (p + 4/a) T (p) -Ti(p + 2/a)) + 2V t T 2 (p + 2/a) ‘

( 26 )

5

SNR p [dB]

Fig. 1. Simulated and high-SNR approximate effective rate against the
transmit SNR p for MISO systems over i.i.d. a-p fading channels ( Nt = 2,
A = 0.5, and p = 2). Moreover, the effective rate of the non-fading channel,
i.e., AWGN channel, is plotted to serve as a benchmark for comparison.

SNR p [dB]

Fig. 2. Simulated and high-SNR approximate effective rate against the
transmit SNR p for MISO systems over i.i.d. a-p fading channels (Nt = 2,
A = 0.5, and a = 4). Moreover, the effective rate of the non-fading channel,
i.e., AWGN channel, is plotted to serve as a benchmark for comparison.

more multipath components and a large value of a accounts for
a larger fading gain from the physical model, respectively. We
also note that the gap between the AWGN and fading channel
curves becomes smaller as a and // get larger. Finally, it is
worth mentioning that the effects of fading parameters on the
effective rate become less pronounced as they increase.

The simulated and \ow-Eb/No approximate effective rate
are depicted against the transmit energy per bit Eb/Ng
for different delay constraints A in Fig. [3 Clearly, the low-
Eb/Ng approximations are sufficiently tight and become exact
at low Et/No values for all the considered scenarios. The
effective rate is a monotonically decreasing function of A ,
which implies that tightening the delay constraints reduces the
effective rate. However, the change of the delay constraint A

Fig. 3. Simulated, Iow-E^/Nq approximate effective rate against the E^/Nq
for MISO systems over i.i.d. a-p fading channels (Nt = 2, a = 2, and
P = 2).

does not affect the minimum Eb/No, which is —1.59 dB in
our case. In addition, the curves in Fig. [3 also show that the
accuracy of \ow-Eb/No approximate solution is improved for
smaller values of the delay constraints.

V. Conclusions

In this paper, we have presented a novel approach to analyze
the effective rate of MISO systems over i.i.d. a -// fading chan¬
nels. The proposed technique is based on the highly accurate
approximation of the sum of i.i.d. a-ft RVs by another a-fi
RV. Novel and analytical expressions of the exact effective rate
of MISO systems over i.i.d. a-/r fading channels have been
derived. Moreover, we have presented closed-form expressions
of the effective rate in the high-SNR regime to gain physical
insights into the impact of system and channel parameters on
the effective rate performance. For example, the effective rate
can be improved by utilizing more transmit antennas as well as
in a propagation environment with larger values of a and p. In
addition, our analysis provides the minimum required transmit
energy per information bit for reliably conveying any non-zero
rate at low SNRs. Finally, numerical results corroborate the
high accuracy of the proposed approximations. Our analytical
results serve as a performance benchmark for our future work
on the performance analysis of the multi-user scenario.

References

[1] G. J. Foschini and M. J. Gans, “On limits of wireless communications
in a fading environment when using multiple antennas,” Wireless Pers.
Commun., vol. 6, no. 3, pp. 311-335, Mar. 1998.

[2] E. Telatar, “Capacity of multi-antenna gaussian channels,” Europ. Trans.
Telecommun., vol. 10, no. 6, pp. 585-595, 1999.

[3] D. Wu and R. Negi, “Effective capacity: A wireless link model for
support of quality of service,” IEEE Trans. Wireless Commun., vol. 2,
no. 4, pp. 630-643, Jul. 2003.

[4] G. Femenias, J. Ramis, and L. Carrasco, “Using two-dimensional
Markov models and the effective-capacity approach for cross-layer de¬
sign in AMC/ARQ-based wireless networks,” IEEE Trans. Veh. Technol.,
vol. 58, no. 8, pp. 4193^1203, Oct. 2009.

6

[5] S.-W. Ahn, H. Wang, and D. Hong, “Throughput-delay tradeoff of
proportional fair scheduling in OFDMA systems,” IEEE Trans. Veh.
Technol., vol. 60, no. 9, pp. 4620^1-626, Nov. 2011.

[6] W. Cheng, X. Zhang, and H. Zhang, “QoS-aware power allocations for
maximizing effective capacity over virtual-MIMO wireless networks,”
IEEE J. Sel. Areas Commun., vol. 31, no. 10, pp. 2043-2057, Oct. 2013.

[7] M. C. Gursoy, “MIMO wireless communications under statistical queue¬
ing constraints,” IEEE Trans. Inf. Theory , vol. 57, no. 9, pp. 5897-5917,
Sep. 2011.

[8] M. Matthaiou, G. Alexandropoulos, H. Ngo, and E. Larsson, “Analytic
framework for the effective rate of MISO fading channels,” IEEE Trans.
Commun., vol. 60, no. 6, pp. 1741-1751, Jun. 2012.

[9] C. Zhong, T. Ratnarajah, K. K. Wong, and M. S. Alouini, “Effective
capacity of correlated MISO channels,” in IEEE Int. Conf. Commun.
(ICC), Kyoto, Japan, Jun. 2011.

[10] C. Zhong, T. Ratnarajah, K.-K. Wong, and M.-S. Alouini, “Effective
capacity of multiple antenna channels: Correlation and keyhole,” IET
Commun., vol. 6, no. 12, pp. 1757-1768, Aug. 2012.

[11] X.-B. Guo, L. Dong, and H. Yang, “Performance analysis for effective
rate of correlated MISO fading channels,” Electron. Lett., vol. 48, no. 24,
pp. 1564-1565, Nov. 2012.

[12] P. K. Chong, S.-E. Yoo, S. H. Kim, and D. Kim, “Wind-blown foliage
and human-induced fading in ground-surface narrowband communica¬
tions at 400 MHz,” IEEE Trans. Veh. Technol., vol. 60, no. 4, pp. 1326—
1336, May 2011.

[13] A. Michalopoulou, A. Alexandridis, K. Peppas, T. Zervos, F. Lazarakis,
K. Dangakis, and D. Kaklamani, “Statistical analysis for on-body spatial
diversity communications at 2.45 GHz,” IEEE Trans. Antennas Propag.,
vol. 60, no. 8, pp. 4014-4019, Aug. 2012.

[14] J. Zhang, M. Matthaiou, Z. Tan, and H. Wang, “Effective rate analysis
of MISO p-p fading channels,” in IEEE Int. Conf. Commun. (ICC),

Budapest, Hungary, Jun. 2013, pp. 5840-5844.

[15] J. Zhang, Z. Tan, H. Wang, Q. Huang, and L. Hanzo, “The effective
throughput of MISO systems over n-p fading channels,” IEEE Trans.
Veh. Technol., vol. 63, no. 2, pp. 943-947, Feb. 2014.

[16] M. D. Yacoub, “The a-p distribution: A physical fading model for the
Stacy distribution,” IEEE Trans. Veh. Technol., vol. 56, no. 1, pp. 27-34,
Jan. 2007.

[17] D. Da Costa, M. D. Yacoub et al., “Highly accurate closed-form
approximations to the sum of a-p variates and applications,” IEEE
Trans. Wireless Commun., vol. 7, no. 9, pp. 3301-3306, Sep. 2008.

[18] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and
Products, 7th ed. San Diego, CA: Academic Press, 2007.

[19] B. D. Carter and M. D. Springer, “The distribution of products, quotients
and powers of independent H-function variates,” SIAM Journal on
Applied Mathematics, vol. 33, no. 4, pp. 542-558, Dec. 1977.

[20] K. P. Peppas, “A simple, accurate approximation to the sum of Gamma-
Gamma variates and applications in MIMO free-space optical systems,”
IEEE Phot. Technol. Lett., vol. 23, no. 13, pp. 839-841, Jul. 2011.

[21] A. P. Prudnikov, I. U. A. Brychkov, and O. I. Marichev, Integrals and
Series, Volume 3: More Special Functions. Gordon and Breach, 1990.

[22] V. Adamchik and O. Marichev, “The algorithm for calculating integrals
of hypergeometric type functions and its realization in REDUCE sys¬
tem,” in Proc. Intern. Conf. Symbolic Algebraic Computation, 1990, pp.
212-224.

[23] Wolfram, “The Wolfram functions site,” Available:
http ://functions. wolfram .com

[24] A. Lozano, A. M. Tulino, and S. Verdu, “Multiple-antenna capacity in
the low-power regime,” IEEE Trans. Inform. Theory, vol. 49, no. 10,
pp. 2527-2544, Oct. 2003.